Further Results on the Efficiency of Markets for Foreign Exchange

s from these conditions cannot be the basis of a test of market efficiency, although it may be helpful for descriptive or forecasting purposes. The second popular technique for testing spot market efficiency has relied on the profitability of simple filter rule trading strategies.7 Some results indicate that small filters would have been profitable for some currencies during the floating rate period. However, there are many factors which cast doubt on the interpretation of these results. First, it is not clear, ex ante, that the size of the filter can be determined which assures or optimizes profits. Second, even a filter rule which earns a profit over a sustained period is likely to report losses during some interim periods. Thus there is an element of riskiness in these trading strategies which is difficult to measure and difficult to compare to some standard model. Data considerations do not allow us to analyze another potential problem related to filter rule strategies. Market-maker quotations are typically valid only for a small and specified volume of contracts and for a limited time span. It is therefore possible that supply and demand elasticities are sufficiently large so that unusual profits would be eliminated quickly after a small volume of trading. This is important if we want to distinguish an inefficient market which permits $10 billion worth of profitable transactions in one hour, versus an inefficient market which eliminates a profit opportunity after $1 million of trade in one minute. Forward-Market Efficiency. Empirical tests of forward market efficiency surveyed in Levich (1979) can be conveniently divided into four groups. First, there are regression tests which estimate models of the form (1) St+n = a + bFt,n + ut